The present invention relates to an output device of a solid-state image device, particularly to an output terminal thereof.
The solid-state image device comprises a directional array of charge coupled devices (CCDs) horizontally or vertically arranged in series, whereby transferred signal charges are detected. Signals to be transferred are received by photodetectors such as photodiode. Exited electric charges by light energy are transferred through the directional array of CCDs, and an output signal is generated with amplification therefrom.
In the CCD, the charges are transferred from a potential well to adjacently potential well that is formed in a semiconductor region or channel region by applying a transfer clock to gates formed of polysilicon, isolated from the semiconductor region by insulating layer thereon, and having a given interval each other or one another.
Recently, a channel-buried-type CCD (BCCD) is widely used, wherein the charge transfer region is formed in a semiconductor region covered with an insulating layer thereon and beneath gate electrodes. The BCCD performs a wide variety of electronic functions such as image sensor, signal delay line, electronics apparatus using shift register, television camera, etc.
Such a BCCD has many problems to be solved in order to improve the charge transmission speed. The reliability thereof is based on the stability of the signal detected at the final output terminal. The circuit and construction concerning the output terminal of a BCCD is disclosed in "SOLID-STATE IMAGE DEVICE" published on March 26, Japan Sho Whoa 61-9975, "IEEE TRANSACTION ON ELECTRON DEVICE" VOL. 36, NO2, PP 360-366, FIGS. 1 and 8, published in February, 1989.
The above two prior construction employ reset transistors at the output terminal. FIGS. 1, 2A and 2B attached to this specification respectively represent a cross sectional view of the prior constructions, equivalent circuit thereof, and the output waveforms thereof.
Referring to FIGS. 1, 2A and 2B, there is formed in a P well 2 on a substrate an N-type layer 3 which embeds a potential forming P+ diffusion region 4 formed under a transfer gate 8, and N+ diffusion regions 5 and 6 constituting the source and drain of a field effect transistor. Namely, the N+ diffusion regions 5 and 6 and a reset gate 11 all together constitute an N-type reset transistor 576. Between the reset gate and transfer gates 8 and 9 is arranged a pass gate 10 connected to internal operating voltage V.sub.DD.
The N+ diffusion region 5 that is the source of the N-type reset transistor 576 serves the output terminal of the BCCD, whose signal is amplified through a source follower amplifier 20 to produce final output V.sub.OUT. The P-type well 2 and the N-type layer 3 of channel transfer region constitute a PN junction diode 25, of which cathode (N-type layer) is connected to the source 5 of the N-type reset transistor 576 and anode (P-type well) to ground voltage, as shown in FIG. 2A.
Also there exists a resistance 12 between the diode 25 and ground voltage. A detection port 53 is between the source 5 of the N-type reset transistor 576 and the cathode 3 of the diode 25. The input current I.sub.IN introduced to the detection port 53 is determined by the potential V.sub.IN of the transferred charges, and the signal applied from the detection port 53 to the source follower amplifier 20 varies with the clamping characteristic of the diode 25 and the on/off operation of the N-type reset transistor 576. The source follower amplifier 20 is a current-mirror-type amplifier comprised of two NMOS transistors 21 and 22 connected to internal voltage terminal V.sub.DD, and of two ground connected transistors 23 and 24. The above described BCCD detection circuit having the diode 25 between the detection port 53 and ground voltage terminal is called a floating-diode amplifier (FDA).
Referring to FIG. 2B, if a transfer clock .phi..sub.H become "high" at time t.sub.1, the transferred charges pass through the N-type layer (or channel region) under the pass gate 10 so as to the N+ diffusion region (or source) 5 of the reset transistor 576. At time t.sub.2 if a reset pulse .phi..sub.RG of about 15 V are applied to the reset gate 11, the reset transistor 576 is turned on so that the output voltage V.sub.OUT of about 15 V is produced through the source follower amplifier 20. Thereafter, at time t.sub.3 if the reset pulse .phi..sub.RG become "low" the electrons formed in the channel region 7 between the N+ diffusion regions 5 and 6 return to the N+ diffusion regions 5 and 6 so that the output voltage V.sub.OUT is dropped. This is called "feedthrough".
Assuming that the quantity of the signal charges is Q.sub.o, the capacitance between the detection port 53 and P-type well 2 (or PN junction capacitance) C, the voltage gain of the source follower amplifier A.sub.V, and the transfer conductance of the amplifying MOS transistor in the source follower amplifier gm, the output voltage gain .DELTA.V is expressed as follows: EQU .DELTA.V=A.sub.V .multidot..DELTA.V.sub.A =(Q.sub.O /C).multidot.gm R/(1+gm R) (1)
wherein .DELTA.V.sub.A indicates the feedthrough level shown in FIG. 2B, and R the value of the resistance 12 shown in FIG. 2A. As a result, that the output voltage V.sub.OUT is dropped due to the feedthrough is caused by the fact that the charges formed in the channel 7 are partly distributed to the parasitic capacitor of the PN junction. Namely, as the PN junction capacitance C is increased, the output voltage gain .DELTA.V is decreased. Moreover, it is advantageous that the quantity Q.sub.o of the signal charges applied to the detection port 53 is great, for the source follower amplifier may critically respond to the general frequency noises if the quantity is small.
Further, in the above circuit, kTC noises (reset noises) are mixed with the output signal because of the switching operation of the reset transistor. The kTC noises are caused by the electron energy fluctuation that is in turn caused by the fact that when the reset pulse kept to be in the high voltage of about 15 V for a brief time is dropped to 0 V, they are switched with a large swing width so that the electrons in the channel 7 return to the N+ diffusion regions 5 and 6 (or source and drain regions). The kTC noises are small in magnitude, but considerably influencing the transferred signal of insufficient quantity. Description concerning the kTC noises (reset noises) are disclosed in the above "SOLID-STATE IMAGE DEVICE", pp 140-145.